* Correct falsely-referenced attribute into an entity which is an obfuscated name
* style: correct syntax for custom attributes
* fix: remove unused import
Co-authored-by: MyAnoneNeko <MyAnoneNeko@users.noreply.github.com>
* added option to include permissive network security config file during build
* added tests for app with existing network config and for app without
* minor fixes for pull 2791
* refactor: slim down test app for network config
* style: remove extra newlines
* refactor: moved network tests to aapt2
* refactor: remove unused exceptions
* test (aapt2): ensure aapt2 is used for net-sec-conf
* fix (cli): block use of net-sec-conf on aapt1
* fix conflict
Co-authored-by: Connor Tumbleson <connor@sourcetoad.com>
Co-authored-by: Connor Tumbleson <connor.tumbleson@gmail.com>
* feat: update internal framework to API 31 (12 Release)
* refactor: rename enums for new overlays/stagedAliases
* fix: introduce basic staged alias support
* feat: update internal framework to API 32 (12L Release)
* fix: add support for overlay spec skipping
* fix: enforce allowable classes during yaml parsing
* fix: rename class to reference escaping nature of strings
* test: assertion for parsing malicious yaml
We take advantage of the fact that baksmali can determine the minimum API version needed for a given collection of opcodes and extract it, then save to mMinSdkVersion, which is only used by smali via buildSourcesSmali.
This change, in addition to my previous change, allows to decompile and recompile JARs without worrying about specifying an API version via -api/--api-level.
As an extra, added webp file extension to NO_COMPRESS_PATTERN, according to AAPT2 standards, and since it's factually never compresses as it's an already compressed format.
https://cs.android.com/android/platform/superproject/+/master:frameworks/base/tools/aapt2/cmd/Link.cpp;l=2328
If the decoding fails and there are UTF-8 decodable bytes before the bytes that couldn't be decoded, then the read index of the original buffer is incremented and those bytes will be missing from the decode result.
Now we create a new buffer and the decoding will start at the original start offset.
issue #2546
